Skip to content
Shyenova

Area 01

Which of your supply chain decisions would change if the evidence behind them were better?

Most supply chains are the fossil record of old decisions. We re-open the ones that still bind, and build only the analytics those decisions need.

8–15%

Total landed cost

Typical reduction where a footprint has not been re-optimised in five years or more.

15–30%

Inventory released

At constant or improved service, where policy has been set by rule of thumb.

5–15 pts

Forecast accuracy

MAPE improvement where no segmented statistical baseline previously existed.

These are the improvement ranges these levers are known to produce in industry where the underlying practice has not been examined recently. They are targets to test against your own baseline, not claims about named past clients.

A supply chain is rarely designed. It accumulates. A warehouse opens because a large customer asked for it, a safety stock rule survives three planning system migrations, a supplier stays sole-sourced because qualifying a second one was never anybody's quarter. Five years later the operation is a set of answers to questions nobody is asking.

This practice works the structural layer: where inventory sits, where nodes are placed, how the plan gets made, who the money goes to and what happens when a link fails. These decisions are unglamorous, they are made rarely, and they set the ceiling on everything the execution teams can achieve afterwards.

They are also, every one of them, decisions, which is why the analytics sits inside this practice rather than beside it. Data foundations, forecasts, optimisation models and the operating model around them are not a separate discipline here. They are how a footprint gets settled with evidence instead of an argument, how a buffer gets sized against observed variability rather than a rule of thumb, and how a plan becomes something the business will commit to. Analytics that changes no decision is a cost centre with good visual design; a structural decision taken without it is usually taken on the loudest opinion in the room.

What this is not

We are not a 3PL, a freight broker, a software reseller or a systems integrator. We place no freight, fill no warehouse, licence no planning platform, hold no vendor partnerships and take no referral fees, and we do not implement what we recommend. We will help write the statement of work, evaluate integrators and govern the programme, and the build is done by somebody whose incentives we do not share. That is the entire reason a recommendation from us is worth reading.

How we help

12 services in this area.

Each is a piece of work we would sign a statement of work for. Most engagements start in one and end up touching two.
01

Network Design

Where to make, where to hold, where to ship from, settled with a cost-to-serve model instead of an argument.

Read more
02

Demand & Inventory Planning

One demand number the business will commit to, and an inventory policy that follows from it rather than from a rule of thumb.

Read more
03

Strategic Sourcing

Category strategy, should-cost models and negotiation preparation that survives contact with a capable supplier.

Read more
04

Logistics & Distribution

Freight cost, warehouse productivity and 3PL performance, measured properly, then fixed at the root.

Read more
05

Supply Chain Resilience

Multi-tier supplier mapping, scenario stress-testing and the buffers worth paying for, priced explicitly.

Read more
06

Control Towers

Control towers, planning system selection and analytics that change a decision, assessed before you sign a licence.

Read more
07

Digital Twin

A validated model of your operation you can run decisions through before committing to them, tested against history, not rendered in 3D.

Read more
08

Data Foundations

The unglamorous layer everything else rests on, pipelines, models, tests and lineage, scoped to a capability rather than to the whole enterprise.

Read more
09

Forecasting

Models measured against a naive benchmark and backtested honestly, including the finding that some of them should not be deployed.

Read more
10

Decision Science

Allocation, scheduling and network problems solved properly, with the constraints that genuinely bind and a model you can interrogate.

Read more
11

Analytics Operating Model

Where analytics should sit, who owns models in production, and how to stop building things nobody uses.

Read more
12

Applied AI

Language models and machine learning put where they measurably beat the alternative, with an evaluation harness proving it before anything ships.

Read more

Artificial intelligence

Where machine learning earns its place, and where it does not

It earns its place wherever there is rich history, a repeated decision and a measurable outcome: promotional and price response, lead-time prediction where supplier behaviour is observable, intermittent demand that classical methods were never designed to cope with, and extraction from unstructured documents. It does not earn it on sparse data, structural breaks, or anything requiring causal reasoning about a situation with no precedent, and much of this discipline is not a learning problem at all. Network design is an optimisation, inventory policy is a formula, a control tower is an alerting problem. Roughly a third of the forecasting work we do concludes that a simpler method wins, and we report that rather than shipping the complicated one anyway.

How we approach applied AI

What an engagement looks like

A representative shape.

Illustrative. This is the shape of work of this type, not an account of a specific client. We do not publish named results, and we would be wary of a firm that publishes them without the client in the room.

12 weeks

  1. 01Situation

    A consumer goods manufacturer was opening a third distribution centre on the strength of a regional sales forecast, with freight cost per unit rising and no cost-to-serve view below channel level.

  2. 02What we did

    Reconstructed eighteen months of shipment data into a true origin–destination model, built cost-to-serve to customer level, and tested the proposed footprint against demand growth, rate shock and node-loss scenarios.

  3. 03Where it landed

    The third site turned out to be robust only under the most optimistic growth case. Redrawing catchments and re-tendering two lanes captured most of the projected saving without the capital, and the site decision was deferred rather than cancelled.

Questions

About supply chain & analytics.

Where do most engagements actually start?

With a symptom rather than a service. Freight cost per unit climbing, inventory and service both deteriorating, a forecast nobody trusts, a reporting estate nobody acts on. The first fortnight is spent establishing which of the several things going wrong is causing the others, and that diagnosis frequently redirects the engagement away from what was originally scoped.

How much data do you need before you can be useful?

Twelve to eighteen months of transactional data (shipments, orders, inventory positions) plus a rate card and a customer master. If it is incomplete or unreliable we say so in the first fortnight rather than modelling on sand. Reconstructing a defensible baseline is part of the work, not a prerequisite for it, and it is frequently the most valuable thing the engagement leaves behind.

We already have a data team. What would you actually add?

Usually the discipline rather than the capacity, and the domain context rather than the technique. Benchmarking against naive methods, backtesting with proper temporal separation, auditing for leakage, building the evaluation harness before the system. These separate a model that survives production from one that disappoints, and capable teams frequently have never had to apply them. The other half is knowing which supply chain question is worth pointing a model at.

Do we need to buy planning, network or analytics software first?

Almost never, and deciding early is a mistake. Process failures reappear inside new software at considerable expense, and a platform bought to answer a one-time question is the wrong order of operations. We fix the cadence, the parameters and the structure in what you already own; if a genuine capability gap remains, the requirement is then specific enough to write a sensible tender against.

Should the analytics sit centrally or with the supply chain team?

It depends on your decision geography, and reference models are unhelpful. Where decisions are similar across units, a centre builds once and deploys widely. Where they differ by market or product, analysts sitting close to the planners have far better context and outperform. Most organisations at scale land on hub and spoke; the failure mode is picking a structure first and rationalising it afterwards.

What if the answer is that nothing much is wrong?

Then we say so, in writing, with the evidence. It has happened. A study confirming that a footprint is robust across scenarios, or that an incumbent forecast is already near the achievable accuracy floor, converts an open argument into a closed decision and frees management attention for the constraint that actually binds.

Start a conversation

Talk to us about supply chain & analytics.

Most useful first conversations are thirty minutes and involve no proposal. Describe the problem you are looking at and we will tell you honestly whether it is one we can help with.

Or write directly to hello@shyenova.com